双自回归滑动平均模型风速预测研究

Wind Speed Prediction Research Based on Double ARMA Models

风电场中风速变化的随机性很强。对随机过程的建模和预测,自回归滑动平均模型(ARMA)具有较好的效果。以张家口尚义风电场实测风速构成时间序列样本,首先通过差分处理将原始风速序列变为平稳随机序列,并确定该序列的描述模型为ARMA(0,4)。用该模型对验证风速序列进行超前一步预测,得到较好的风速预测效果。为进一步提高预测的精度,对样本序列风速预测的残差再次采用ARMA模型进行建模和预测,并用预测残差来修正风速预测值。对实际风速序列进行预测和验证,结果表明本文提出的双ARMA模型预测可以显著提高风速预测准确性。

In the wind farm, the wind speed has the character of great random. As to the modeling and prediction of random process, the ARMA model has good application. The wind speed data from the Shangyi Wind Farm in Zhangjiakou City are employed as samples. The original data are transferred to the smooth random sequence through difference operation. The smooth random sequence can be identified as the ARAM（4, 0） model. Tbe predictive accuracy is satisfactory by using the ARMA model to predict the one-step iead wind speed .In order to improve the predictive accuracy, the ARMA model is used to model the residual errors of wind precition, and the prediction of residual errors is used to rectify the wind speed prediction. Simulations prove that the double ARMA modeling method can effectively Hnprove the prediction accuracy of wind speed.